 OK, great. Well, thank you very much. Thanks for coming. I just thought I'd start off by saying that this is exciting for us because this is truly a Berkman collaboration. So Erin and I were both fellows a couple of years ago at Berkman. And that's when we decided to work together on this project. So it's really a Berkman project. And in true collaborative form, we're going to present it together. You should feel free to interrupt us if we say something where you'd like clarification. And yeah, so Erin will start with motivation for the project, research questions. I'll talk a little bit about the data and some of the findings, and then I'll pass it back to Erin. That was good. Thanks, Esther. OK, so the place this paper started was really with these questions about what the Obama campaign meant for political engagement, and particularly the relationship between the internet and political engagement among young people. And so Esther has this amazing survey data about a lot of college students in Chicago, which we'll tell you more about in a moment. And so what we were really interested in was to come up with a way to test this idea, to do a pretty strong test for the idea that the Obama campaign had really changed the game in terms of young people's online engagement. And so we thought, what better place to look at this than a bunch of Chicago college students? And in particular, what we wanted to see was whether internet use patterns matter, which is something that previous research on young people's political engagement hadn't really included data about. So Esther will tell you more about what our data has to say about that in a minute. So we've read a lot about what previous research had found on this, and the main findings are probably familiar to a lot of you in some ways in various forms. The big story of young voter engagement over the past 50 years in the US is generally summarized as a pretty sad one. And so far as there's been generally a downward trend in terms of young voters becoming increasingly disaffected, that may have been changing in the past eight years. But long-term trend seems to be generally downwards. And past research in terms of what the internet has meant for this and what people's use of the internet has meant in relation to this trend has seemed to suggest that the internet is usually a weapon of the strong. Or in other words, the people who have greater social capital, greater resources, come from privileged backgrounds, tend to make more and better use of the internet. And that the relationship between the internet use and political engagement only holds for those people from more privileged backgrounds. So the other piece of the previous research has been this kind of argument about whether the internet impacts political engagement directly or whether it affects people's political engagement uniformly or whether there's some other more complicated process. So what factors intervene? What processes matter more than others? And there hasn't been a lot of agreement on that in terms of the social science sources that we looked at. So that's where we started. And we decided to think about political engagement in terms of four specific outcomes. So I'm just going to walk through these a little bit as sort of a preface to what the models that we actually ran. The first is what we call online political cognitive engagement. And what that translates into is the extent to which you pay attention to or participate in political information seeking and political communication on the internet. The second one is civic engagement. And we're really interested here in sort of traditional forms of civic engagement. So volunteer organizations, participation in the community, things that people don't necessarily do online but that they may do online. But sort of the classic, if you've read Bowling Alone, Bob Putnam's book on civic engagement, social capital, you'll have an idea of what we were thinking of in this area. Of course, we were interested in voting as an outcome as well. That's the sort of obvious one. And we were also interested in political action more broadly defined. So Esther will have more about what specifically went into that measure. But as you can see from the arrows here, the other piece of this is that we were using, the way we were running the models is we were trying to incorporate these in a cumulative way. So implicitly what that means is we were assuming that certain things about people's attributes might affect the extent to which might associate with the extent to which they engage politically online, which in turn might associate with the extent to which they are civically engaged, which in turn might associate with the extent to which they vote or engage in other forms of political action. So that's the kind of 10,000-foot overview of what we were doing. And that's what Esther for the data piece. OK, so now I'll tell you about the data that we're using. So some of you know about my work and know that I've done a lot of data collection at the University of Illinois, Chicago. So I'll just say a few words about why there. I've never been affiliated with that institution myself. So it's not about just simply surveying my students. The reason that I've been studying students at UIC is that it's an extremely diverse group of students. It's one of the most ethnically diverse campuses in the nation when it comes to research universities. And so that is helpful when one is interested in the kinds of questions that drive my work, which have to do with social inequality with respect to internet uses. So that's why we use the data set from UIC. What we do is, or what I did with my research team at Northwestern, is going to a class at UIC that everybody is required to take on campus every first year. So this way we're not biasing towards some type of class that people might be interested in taking. And I'm very grateful for the first year writing program for working with me on this. It is a paper-pencil survey, which is less and less common. A lot of surveys are done online these days. The reason that I continue to use paper-pencil surveys is that a lot of the outcomes of interest have to do with internet uses. And so I don't want to mix the mode of data collection with the research questions of interest. So that way I want to remove questions such as who has access to the internet who's more comfortable taking these surveys online from the mode of data collection. So that's why it's paper-pencil survey. It might sound simple. One course, paper-pencil, going to class. It's actually 86 different sections. So it's a lot of work to collect all this data. I just wanted to clarify that. So these data come from the spring of 2009, where we had over 1,000 first years take the survey with quite a high response rate. We today focus on just 1,001 of those students who were actually eligible to vote, because we're not really interested in the students right now who are not eligible to vote, since that's one of our important outcomes. OK, so we'll tell you a little bit here about what the sample looks like. So oh, sorry. Are those students who are eligible to vote in 2008? Or are they ones who are eligible to vote starting in 2009? This was for the 08 elections. So the majority of the students are 18 and 19. They're young. We have a bit more females than males in the sample. Parental education is the variable we use as a proxy for socioeconomic status. You can see that almost half of the sample is first generation college students. Again, it's a pretty diverse sample and also quite diverse on race and ethnicity. That said, we do want to remind you that this is a specific population. So we don't want to suggest that our findings are generalizable to all of the US. But we think that it's a very interesting population, especially given that it's Obama's hometown. So 62% of the students in the sample voted in the presidential elections, which compares very well with the figures nationally for that age group and education level. So this is actually, in that sense, quite representative. Very similar numbers. We collected data on what we call political capital with the literature suggests our variables that might matter when it comes to political engagement, just so we can control for these. So one has to do with partisanship. We have information on people's level of political interest and also political knowledge. And we relied on pretty standard ways to measure these. If people are interested in more detail, we can get into that. Would range be like, thank you, partisanship problem. OK, so I guess I'll get into how we project these. So the partisanship question specifically came from, or variable, came from a question that actually asked people about their ideology from very liberal all the way to very conservative, where we coded very liberal and very conservative as three on the partisanship, liberal or conservative as two and middle of the road as one. So we recoded that question. OK, this is definitely the wired generation. So people talk about this cohort as people who've grown up with technology. And indeed, we do find that when it comes to very basic levels of internet use, they've really been online for many years. And so there are no questions of a core digital divide when it comes to this population. So that's something that we've pretty much controlled for by looking at this group. However, when it comes to more nuanced measures of internet use, there is some variability in the group. So these are measures that, again, the literature would suggest might matter for internet use, some of it. Measures that I've introduced, for example, people's level of skill in using the internet. Just to clarify, the range here refers to what the options were in the survey questions. Those are not the minimum maximum values of the responses. So for example, everyone's been online already by the time of the study. OK, so onto our important question of what we call online political cognitive engagement. So the way we operationalized this was to look at a few questions that have to do specifically with engaging with political types of content online. And we look at two types. One has to do with blogs in particular. Are you reading blogs that relate to politics? But it's actually a pretty broad category, including economics, law policy. That's how those were the words used in the survey. Are you commenting on these types of blogs? Also looking at whether you visit websites with these topics and whether you engage with content in a more active way of this sort. As in, do you forward? Do you post? Do you respond to this kind of content? And so these are the figures, which we show you because this is such a central variable in the study. So note that most of the numbers under Never are quite high actually, right? So we have a lot of people in the sample, a lot of these young adults who are not really engaging with this type of material online. And especially when it comes to the more active form of commenting, of engaging with the material by forwarding, those numbers are very high, meaning that about it's really just one fifth at maximum who are doing any of that work. Mary. I think if you said it, did you see what you meant by political or economic law policy? No, this was just, this is what it said, yeah. So what we then did was we created a summary variable of this information. That's just a binary variable where if you did any of these things weekly or more often, that's where you get a one on that zero one binary variable, which is about half the sample. So then we ran a logistic regression with information about people's background, political capital measures, and some of the internet measures to see what relates to online political cognitive engagement. And instead of showing you the raw numbers from the regression, we came up with this representation. Let's see if it works. Basically, what we found that women are less likely to engage with this kind of material online, race and ethnicity, as well as parental education, which is our proxy for socioeconomic status, do not seem to matter, don't relate. Interest in politics does. This is not shocking. You're interested in politics. You're more likely to be engaging with this kind of material on the web. And also, how often you're online and your skill are positively related to online political cognitive engagement. Next, we looked at civic engagement as Erin mentioned earlier. We rely on fairly traditional measures of these. These are specifically what we asked on the survey and what we use data on. And this is how prevalent these types of activities engagements were in the sample. And what we did here was we added up the number of these that you did, either ever or in the cases where you see in parentheses a few times a week or more. And so here we have an interval outcome from 0 to 4. Yes? Fact one, did you control women and internet skill together? Well, it's a regression. So it's all in the model at the same time. Is that what you're asking? Whether the negative effect was independent or dependent on internet skill kind of a problem. So actually what we did with the models was we first ran a model with just the background variables. And then we added the internet variables. And I remember for sure that when we add the internet variables, that is a negative relationship. So controlling for skill. Yes. So now the outcome is this 0 to 4 interval variable of civic engagement, your level of civic engagement. And here gender doesn't seem to matter. However, Asian-Americans seem to be more engaged than whites. Parental education matters in the sense that those whose parents have higher levels of education engage more. Civically, interest in politics is related. OPC engagement refers to our online political cognitive engagement measure. So there's a positive relationship there with engaging online in political types of content and being civically engaged, whether online or offline. And internet experiences, when it comes to autonomy, your use of social network sites and skill are all positively related. Nicole. Could you talk a little bit about why you decided to focus on blogs for the political engagement as opposed to newsrooms or chat rooms? So actually the latter is included in the other. So it's four measures, two types. One has to do specifically with blogs. And the other is a much more general question that just said, do you go online to look at these types of content? Which is actually inclusive of it could be news groups, it could be just a newspaper website, anything. So that's very inclusive. We specifically just highlighted blogs as a separate question, just to focus in on that because it's gotten so much attention. But that's like an add-on. It's not replacing other types of. OK, so that measure there. This one measure, it's all those four. So reading, looking at blogs, other types of political content, plus commenting on blogs, plus forwarding an article you read in the New York Times, all of that is supposed to be included. Yes. Yeah, another question? No, no, not at all. No, nothing. Good question. Thank you for asking. Sorry. That's just trying to fit it all in. Over a semester by the time we are collecting the data. So things that college may afford them to do, they could be doing already. That said, I do think one area that would be really intriguing to follow up on here is what it is exactly that's going on on social network sites that they're doing. I didn't specify what those data were about, but basically that variable is simply do you check people's status updates and do you post status updates to yourself with some regularity? That's a pretty crude measure of SNS use for our purposes. It would be great if we had some more nuance there. SNS use specifically for political types of purposes, for example, that would be neat to have. We've had lots of hands up. OK, thank you. I might be misunderstanding your final simplified version of your results, but it seems somewhat paradoxical that internet use doesn't correlate to voting, but internet use does correlate with political action. Or should we assume that voting correlates with political action, or is that a false assumption, or is the data more nuanced in a way that you can't simply make a chain rule and come out with a paradox? So they do relate. A political action of voting are related. OK, so they're related in the context of it. Am I correct in understanding somewhat of a paradox here, or is that just an artifact of how the data is filtered? It's nothing to do with what we did with the data. I think it might just be a paradox in the way that political engagement and action work among American young people. I mean, I think voting is, at least qualitatively, a different kind of engagement in the United States than signing petitions, donating, volunteering for campaigns. And I think part of what we find is that it's reflected in so far as the kinds of people who tend to vote more look a little different than the kinds of people who do those other things more. So I'm not sure it's a paradox, strictly speaking. But it's definitely interesting to disaggregate them and think about them separately, I think, for exactly this reason. And it suggests that some of the factors that drive whether people vote or not may not necessarily overlap with the factors to determine whether people are engaging in a broader sense. So I'm hesitant to draw it. I don't want to extrapolate from these to claim that any of this was clearly causal in any way. But I think that it's suggestive in that regard. So that's the, yeah. We should mention that the R square of these models is around, I think the best fit is what explains 12% of the variance. Yeah, it's not humongous. There are lots of things that we're not including in the models that we have no data on that matter. So we're just capturing a relatively small portion of the variance to be sure most internet use research that's about as much as you're going to get. So in that sense, we're in line with research. But there's obviously other things going on, too, that we're just not capturing. Yes? I'll follow up on this question about the theme. When it goes back to what is eligibility, would a student be considered eligible or self-identifies eligible if they had moved from out of state? Oh, you know what? This is a point I had meant to mention. I forgot. The vast majority of the students in the sample are from Chicago land, specifically from the Chicago area. We actually, we haven't included it here. We have the data. Almost everybody is from Chicago land, just a handful horn. So that's actually not really a concern with this data set. That said, the question was about the presidential election. So they would say they're eligible. It could be from people who are more likely to vote or not when they're away. And that's a very legitimate question, which is why I should have mentioned that almost everybody in the sample is from Illinois. Yes? I'm just wondering how active is the democratic public? No idea. So good question, but I don't know. Yeah. Oh, sorry. The question was how active the Democratic Club is at the university. Yeah, or how various party groups might or might not be. No information on that. I know there's at least one study that was done in Northwestern about sort of get out the vote outreach efforts by specific groups and their effect on the student body there and their participation rates. But we definitely don't have that information for this sample. So it would be interesting to look and see whether there are findings in that area. And that work by Michael Pechkin was actually, especially focused on the idea that at Northwestern, a lot of the students are from out of state. It's not trying to get people registered, but it wouldn't be that easy. But again, with our population, it's their local. So it's that there isn't that extra hurdle. Yes. You said you had later data about the financial rates. Did you look at any other voting, like smaller level local voting, or are you looking at the larger level voting? I have no data on other types of voting. Yeah, and so far with this, we've only looked at the 2009. Yeah, we haven't actually looked at the results. Yes, you have a question here? Yeah. Yeah. So there was a study that came out recently showing that something like 85% of contributors to Wikipedia were male. And I was wondering if you have looked into, since you do collect data on gender, if you saw anything similar related to online political participation in your sample set? So I mean, we showed the. Yeah, I don't remember here. We can go back to it. Here we just have the models. We don't have the breakdown by gender. I could tell you that Wikipedia contributions, I actually have very nuanced data on that. Esther has done wonderful research on exactly this topic, which is elsewhere. And in particular, in the 2010 data set, I have very detailed information separately. Have you edited? Have you started a new entry? Have you looked at the history? Like I have about six or seven different measures. It's unbelievably gendered. I mean, it's the data are almost even more extreme than what you're quoting. The Wikipedia survey. I could even pull up the, no, I have a slide. Afterwards, I'll show you, I have a slide that actually has those data, and it's remarkable. I mean, we're talking extremely large differences. From the 2009 data set, I remember the figures were just generally asking, have you edited a Wikipedia entry? 29% of the men said they had and 8% of the women said they had. So that's a huge difference. But when we look at the specifics of edit versus add versus all those things, there's some even larger differences. So yeah, those are huge. Now, as to political, I mean, I guess on these slides, we don't have the specific breakdown. But gendered, I mean, but you can see here that in this model gender, even when you control for all these other factors, women are still doing it less. So, yeah. Is it as drastic a difference? No, it's not as drastic, no. I don't remember up hand the magnitude of it. Those contribution types of activities from my prior research, and actually some of it, I think I talked about in my last Berkman talk. They're always gendered that women are contributing less. I'm pretty much everything I've looked at. And one of the critiques I got at that point, I looked at contributing to Wikipedia, I looked at posting videos, I looked at blogging, creating quizzes. Maybe that was the one that wasn't, there was no gender difference. That's right, that's right. As in like Facebook quizzes, probably. And then the one critique I kept getting was, oh, but you didn't ask about fan fiction. So in 2010, I asked about fan fiction and women are still reporting less for that too. Now, and then you can start asking, well, is there some misreporting here? I can't really come up, I mean, as a survey researcher with a lot of experience, using survey, I can't really come up with a reason why systematically that would be misreported. So if people have ideas, let me know, but I haven't been able to say that that, there's like a reason that that would be misreported. And that was, you know, the biased way. Come on. You're using these different measures. Is there, how did it compare in terms of receptive use versus productive use? As in, you know, your concept is online political cognitive engagement, so does that include receptive use as in, you know, reading? Yes. It's inclusive of both in this case. I don't remember offhand. I'm sure that we looked at that, but in just a descriptive way at some point. Yeah, where's the, here's the, I mean, we could, if you hang around for a few minutes afterwards, we can check the crosstab, but I mean, I don't know offhand what it was. It's production. Specifically production. Yeah, right. I mean, I can say, so specifically for Wikipedia, in terms of have you read a Wikipedia entry, like 99% for both men and women. So that's universal. For Wikipedia in particular, for these measures, I don't remember right now what the crosstabs are. It may be that men do a little bit more, but I don't recall major variation there. David. I understand that you took partisanship as a single measure of strength, not of where you are, but if you put back in, I don't know if you have the data, but if you put back in left versus right. Ideology. Yeah, does it make a difference? Have we done that? I don't think we did do that. I think we... In part because I think like the previous work in this area, I mean, the motivation for doing it the way we did it, is just the previous work in this area has been overwhelming that it's partisanship, not ideology that matters, but in this particular case, I think that you raise a good point that given the geography and the population involved, it would be worth going back. So yeah, we haven't done it yet. Something we could do. It's a great idea. He didn't say something about the Obama effect. Absolutely. Yeah, we're not always energizing the right as well, so... I think that's the same question, but I was wondering in your data, did you see the people who identified as very passionately one thing as opposed to another voted new? So that's the partisanship variable? Yes. I think so. We do control for that. No, it's the question... If we go back to the model for voting, if we go to the voting outcome model? Yes. So, we're partisan. That's not a shock either, but you found that we're partisan. The people who are more... So the plus sign next to political capital and the fact that partisanship is listed there means that people who are more partisan were more likely to vote. And was there any difference between men and women now? So the idea is that we're including information about all these variables at the same time in the model. So holding one constant, the other is significant? Yeah, the way to read it is sort of like, all other things being equal, women were more likely to vote than... were like more likely to have voted according to this data than men. So if you take all other things being equal at the equal level of partisanship, the women are more likely to vote than the men. Right. Yeah, that's sort of the way to read these. Although you have to be careful with that more likely piece because that suggests that it's linear and causal. Yeah. And that's not... It's complicated. That's a variable. Yeah, yeah, yeah. Yes. Just to go further, drill a little bit down here at this point, women vote as a percentage more often than men. Nationally, I think it's... Nationally, that's the case, right? So is there then even less effect on women? So if you have women who are voting more and you're not seeing a correlation then, is there even less poll on women? Because they're a larger... I don't know how we would pull that out. Yeah, I mean, these wouldn't let us... Yeah, this kind of method wouldn't do that because the mechanics of it don't... Like, yeah, don't let us look at that. It would be interesting to try to figure out if there's a way to pull that out. I'm not sure what it would look like. Yeah, maybe it would just be an interaction. Yeah, we could try it. Just because if you think about what you're saying, if you sort of step back on what you're saying, right, you're saying that men produce more content, engage with more content, but women still vote more. Right, but in fact, engaging with the content here doesn't matter. So, yeah, we're controlling for that. That doesn't matter. So maybe if we interacted with those, some of the online variables... Yeah, it would be interesting just to try it with... Yeah, it would be worth probably going back... If we were to do this, it would be worth going back and looking more at the descriptives of the population to pull out plausible stories like that and then go in and specifically look at those interactions, which I don't know if we did very much... I don't think we did very much of. We first generated this, yeah. So, that's a great point. We were very much led by the literature. We just wanted to test what the literature said and see how can we measure these concepts and then how does this compare? But that's a great point. That's a good one to think about, so we should go back to that one. Rob. So, two questions. One is, of all the things you didn't measure, what's the thing you really wish you had to measure and the second one is getting back to the voting and political participation thing. So, I understand the logic of having different determinants of civic engagement and political engagement being driven by different things, but my prior assumption would have been that political engagement and voting would be more or less linear. It would be on the same trajectory and that the determinants of the both would be similar as well. Do you have a theory or a hypothesis as to why they're not the same or? I mean, to go backwards. I think the second part of your question is actually, goes back to the question moment, too, which is, and I mean, beyond the fact that I think they are qualitatively different in American society, I don't have a good story there. I don't know if you have more to... I mean, they're not perfectly correlated, so there's definitely, in terms of looking at the correlations between those two, they're not, it doesn't line up exactly. It's definitely not something that is sort of there in the... Even if, in a certain way, we can sort of say from a high level, political engagement of these different kinds should be created more or less equally, it's apparently not. So that's sort of the best I can do, unfortunately. I don't know beyond that. In terms of measures that we wish we had, do you have any that... Well, I mean, I'm actually pretty proud of this data set and how much we do have information about it. It's pretty amazing, yeah, yeah. I think with the internet variables, I'd love to have more detail there. Like, as I said, I would love to know what's going on on social network sites that matters. I think it'd be interesting to have maybe more, although I do think the OPC, the online political cognitive engagement does actually capture a lot, so you could have more nuance, but since in some cases it doesn't matter, I don't know if it would help. But I do think that it would be interesting to see more of the nuances there, especially with some of these newer technologies. But do you have thoughts of other people that we'd love to hear, you know? The other question I was gonna ask is, do people look at the impact of social networks on voting behavior in political? There's been some other work on that, yeah. As in just like when your networks are voting, and I mean, I'm sure there's literature on this. I don't think we really looked at that here, although, wait.